Reproduced from the CCRexplorers archives:

Cells are relatively simple in their operation. Oxygen diffuses through a membrane and the gas contacts the sensing electrode and the base solution. This creates a reaction at the wet surface of the electrode; this reaction consumes the electrode.

The oxygen sensor acts as a current source, so the voltage measurement must be carried out over a load resistor.

All oxygen sensors are self-powered, diffusion limited, metal-air battery type comprising an anode, electrolyte and an air cathode as shown above.

At the cathode oxygen is reduced to hydroxyl ions according to the equation:

O2 + 2H2O + 4e- ===> 4OH-

The hydroxyl ions in turn oxidise the metal anode as follows:

2Pb + 4OH- ===> 2PbO + 2H2O + 4e-

Overall the cell reaction may be represented as:

2Pb + O2 ===> 2PbO

The oxygen sensors used are current generators, and the current is proportional to the rate of oxygen consumption (Faraday’s Law). This current can be measured by connecting a resistor across the output terminals to produce a voltage signal.

The anode is seldom a simple chunk of lead because the surface must be accessible for contact with oxygen molecules before the electrochemical reaction can occur. The anode usually consists of a form of the metal that provides a good surface area-to-volume ratio, like a granule or wool.

Other important design features are the temperature compensating thermistors, internal membranes, spacers, wicks, and current collectors. Numerous variations on these general design features are used to optimize the performance of an oxygen sensor for use in a specific product.

One of the most important design constraints is the build-up of lead oxide that develops in the sensor capsule over the life of the sensor. As lead is converted into lead oxide, an increasing fraction of the volume of the sensor capsule is occupied by solid material. If the sensor design does not include a provision for this increase, it will eventually rupture, possibly leading to leakage of electrolyte into the loop or electronics pod. Nasty!

The lead wool or granules forming the anode in the oxygen sensor is consumed over time. When all available surface area of the lead (Pb) anode is converted to lead oxide (PbO), electrochemical activity ceases and current output falls to zero. At this point, the sensor must be replaced. Most fuel-cell sensors are designed to last no more than one to two years.
In rebreathers, even when the rebreather is not being used, the oxygen sensor continues to generate current and use itself up.

The temperature of the atmosphere influences the output from fuel-cell oxygen sensors. The warmer the atmosphere, the faster the electrochemical reaction proceeds. For this reason, oxygen sensors usually include a temperature-compensating load resistor to hold current output steady. More on this later on in the article…
Cold temperatures can be a major factor that limits performance. The freezing temperature of electrolyte mixtures used in some oxygen sensors can be as high as minus 20 degC (-5 degF). Once the electrolyte freezes, electrical output falls to zero. The gel and liquid electrolyte mixtures used in more recent sensor designs, however, show much better cold temperature performance.

The partial pressure of oxygen (or pO2) is that fraction of the total pressure due to oxygen. Partial pressure oxygen sensors rely on the partial pressure of oxygen to drive molecules through the diffusion barrier into the sensor.
As long as pO2 remains constant, current output is a reliable indicator of oxygen concentration. Shifts in barometric pressure, altitude, or other conditions that affect atmospheric pressure will cause a systemic change in sensor output.
Consider a sensor calibrated at sea level where atmospheric pressure is 760 mm Hg. Now consider the same sensor at an elevation of 3,000 m (10,000 ft). Although at both elevations the air contains 20.9 percent oxygen, at 10,000 feet the total atmospheric pressure is only 514 mm Hg, and the partial pressure for O2 is only about 107 mm Hg. Because there is almost one-third less force driving oxygen molecules through the diffusion barrier into the sensor, current output is significantly lower.

Mechanisms of Oxygen Sensor Failure


Oxygen sensors may be affected by prolonged exposure to acid gases, such as carbon dioxide. Most oxygen sensors should not be used continuously in atmospheres containing more than 25 percent CO2. In some cases, prolonged exposure to the acid gas damages the basic (alkaline) sensor electrolyte. In other situations, high concentrations of acid gas produce a current flux that alters the normal expected output of the sensor at a given concentration of oxygen.

As indicated previously, the working electrode in a fuel-cell sensor is consumed over time. When all available surface area of the lead (Pb) anode is converted to lead oxide (PbO2), electrochemical activity ceases and current output falls to zero. This is the type of sensor failure most frequently experienced by users.

Prolonged and Incorrect Storage

The Potassium Hydroxide (KOH) electrolyte contains fluid that evaporates over time when exposed to air. If the sensor dries out the output will drop significantly or results in a zero output from the sensor.
Packaging of the sensor is important to prevent this and also for oxygen exposure. The packaging of sensors in bags would allow you to assume that this could not happen, however the packaging in some sensors is gas permeable and allows oxygen to permeate. A non-permeable packaging would extend the life of the sensor but would also require the sensor to wake up after storage.

Fail High

When a cell fails high, it is invariably a result of a manufacturing fault or mechanical damage. Damage to the membrane or poor manufacturing can lead to the loss of electrolyte, this can cause the output to rise erratically and fail low within a short period of time.
The increasing cell output is transferred to the PO2 controller which responds by stopping injecting oxygen into the loop Result, HYPOXIA.

Current Limiting

There are two reasons commonly found for a cell reaching its current limit. Either the cell has reached the normal end of its life due to the consumption of the lead anode or the cell has limited prematurely due to poor manufacture. When a cell reaches its current limit, its ability to monitor oxygen concentrations high levels are reduced. The rebreather assumes there is insufficient oxygen in the loop and injects to reach a set-point the cell will be unlikely to achieve resulting in HYPEROXIA.

Non Linear, Slow Response and Drift.

The electrolyte is a very aggressive material and reacts well with a number of substances. If the KOH is contaminated by any organics, the result would usually be a reduction in the output of the cell. Gas bubbles in the electrolyte are an additional concern that will create instability or inaccuracy across its range, especially when the sensor is moved or rotated mechanically. This is a particularly
Insidious failure mode for a rebreather, as although the sensor may pass calibration, the system will not track the actual gas content of the loop.



Humidity is a subject that few divers deal with on a continuous basis, and it is easy to misunderstand the various humidity scales. Dewpoint is a well used and useful scale of humidity: it is an indication of absolute humidity and independent of the temperature of a gas.

Trimix contains a mixture of gases, mainly Helium, nitrogen, oxygen, and varying amounts of water vapor  The amount of a particular gas present may be expressed in terms of a partial pressure. The sum of the partial pressures exerted by each of the gas mixture components equals the total pressure. For Trimix:

Helium + Oxygen + Nitrogen + Water Vapor = Total Pressure

The wetter the air, the higher the partial pressure water vapor exerts. Warm gas is able to hold more water vapor than cool air. This can be likened to a sponge – the warmer the gas, the larger the sponge.

As temperature increases, more water is able to be held in the vapor phase. Dewpoint is defined as the temperature to which the gas would need to be cooled in order for condensation to begin. If the gas is cooled further, more water will condense – the temperature and the dewpoint are the same, and the gas is still saturated.

So it should be noted that water vapor has a partial pressure and makes up part of the breathing, and therefore the sensed gas. Rebreather control systems only measure the oxygen content of a gas for PO2 control, however integrated controllers with decompression rarely make adjustments for vapor in the assumed inert gas fractions. Such inaccuracies can only be calculated if the gas content is known for that particular rebreather as the oxygen sensors being positioned in different locations in a loop will vary vastly in supersaturation. Such accuracy in gas content would only be of benefit where such data is critical, modern decompression devices have either built in conservatism or user adjustable conservatism that can be manipulated to take into account those unquantified variables.

Pressure Dewpoint

As a diver, it is important to know that dewpoint also changes with pressure. With an increase in total pressure, each partial pressure is also increased: an increase in total pressure causes a shift in dewpoint. This increase is like squeezing a sponge – gas is unable to hold as much water vapor under pressure.

If the total pressure is doubled, the vapor pressure due to water vapor is doubled. The result is that the dewpoint shifts. In rebreathers, where gas at atmospheric pressure is compressed to 7 Bar or more, the vapor pressure rises a long way above the saturation curve, and condensation is observed.

To give you some perspective, if the loop gas you breathe is at 35 degC on the surface, at 60m the loop gas would have to be at 76 degC in order to maintain the same dewpoint.

Moisture and Sub Systems

It is therefore important to accept that condensation will occur and there is little we can do about it. What we can try to do, is prevent that condensation from causing problems with electronics and sensing systems.

Condensation is not an immediate problem; it is a potential problem in respect to the sub systems. Pure distilled water conducts little electricity; it is the impurities in the water which gives it the conductive properties. When we assume that there will be organic components as well as scrubber dust and other debris, we should assume some conductivity. If saline liquid were to make it in the loop, then we can be assured of it.

There are two main aspects of those systems we are concerned about, the electronics including power supplies and the oxygen cells.

Electronics Systems

From our earlier conclusions, it is pretty obvious that the water vapor is present anywhere there is gas from the loop. It is possible to mitigate the vapor by segregating the electronic components and using water absorbent membranes or cartridges at the interface but this has a disadvantage of being of limited life before it too is saturated and requiring post dive maintenance.

The only other way is to separate the electronics from the loop entirely in its own housing. Unless the electronics are in 1ATM housing, then there are equivalent levels of vapor in the electronics or battery housing regardless of their physical construction.

Electronics on a rebreather system are varied and should be considered from the integrated controllers to the cables interconnecting subsystems. Most electronics circuits in such equipment are potted or encapsulated in a hydrophobic or displacing substance such as a polyurethane two pack. Cables and their termination are a weak point in that they must, at some point, be exposed to the gas in question. This leads to some gas permeating inside the cable sheath and causing corrosion, or even worse, voltage leaks which we describe as Creepage.


Galvanic-assisted corrosion is of particular concern, given the fact that solder joints are comprised of different metals or alloys that are in contact with one-another. Add to this a high oxygen content environment and salts and any metals are at particular risk. To many, this phenomena is countered by “regular” maintenance which usually means “replace when bad”. Whereas it is just as simple to design and construct using principles to avoid this corrosion in the first place. Polyeurothane two pack compounds which are not loaded with fillers are relatively cheap and encapsulating joints is a simple and effective method. Let me explain, soldered joints in rebreathers can corrode by >0.15mm per year when inadequately protected. Oh, and before you ask, standard heat shrink is not protection.

Inevitably, when cables and conductors are inadequately protected, the pressures exerted on the system promote migration of oxidizing agents up the cable through the gaps between conductor strands. When considering the low voltages involved in closed loop control systems in rebreathers, we can see how design criteria can be seen as a critical function of its design, not a secondary consideration of assembly. Marine cables are now commonly available that minimize or prevent this occurrence.

Oxygen Cells

Oxygen Cells rely on a very small output voltage that operate between an effective 8-150mV output. Creepage acting on a system with such a sensitive instrument voltage range is significant.

If water vapor were to condense in the loop and form moisture, which it will, then that moisture would first collect on surfaces with a lower temperature than the gas. Eventually, almost all the surfaces will obtain some moisture and ignoring the effects of wicking the liquid will attach to most surfaces. The level of conductivity of the liquid determines the level of detriment on the system.

An SMB or coaxial cell has a dielectric separation of about 5mm which is fairly good. This is due to the labyrinth pattern of the engagement of the male and female parts of a plug/socket.

A Molex plug does not perform as well as the separation is only the distance between the pins, a couple of millimeters.

An SMB cable socket still has a cavity where the conductor is soldered to the centre pin and this should be potted with polyurethane to prevent any problems.

The cells themselves contain electronic components. This includes a resistor network and a thermistor or similar device to effect the temperature compensation. These components inside the cell should be encapsulated or coated in a conformal coating.

The important part of the cell in respect to oxygen sensing is the face. Water condensing on the face of a sensor will have the effect of slowing down the response time of a sensor and although humidity 0-99% R.H. ( non condensing) has a minimal effect on the sensor the water vapor in the humidified gas will appear as a lower than expected Oxygen output.
So conclude that in a rebreather, calibration of a cell with any occlusion on the face can lead to an incorrect correction factor being applied to the control system.

Remember this; the response time of a cell is a function of two parameters:

  • The Membrane –  This is fixed at manufacture and stays constant throughout the sensor life
  • Henry’s Law – The difference in pressure between outside and inside the sensor.

They are the facts governed by the physics of a sensor. However external effects can alter what happens. Such as water on the membrane and sensor age.

Near the end of the sensor life the anode surface is much smaller so response time must increase (probably a very small effect until the last hours of the sensor). Measuring response time has no purpose as it is less enlightening over measuring a sensors output at high PO2.

Rebreather Design Influences

Reducing the amount of water vapor in contact with an oxygen cells face would be a worthwhile endeavor  The laws of physics tell us that we have few variables to play with. Pressure is not under our control as the loop is ambient with the pressure exerted at depth. By designing a system right, we can minimize the effects at the cell face. Let’s look at that.

Imagine how a dehumidifier works. Air in a dehumidifier passes over a series of cooling coils (the evaporator) and then over a set of heating coils (the condenser). It then goes back into the room as drier air with its temperature elevated. Sound familiar?

The scrubber produces water during an exothermic reaction, 99% RH can be expected in the inspiration gas. The water vapour pressure will however depend on the gas temperature. If the gas in the loop was passed over a cold surface before the scrubber, and then the gas from the scrubber kept at an elevated temperature until it is passed the cell faces, then you would have the effect of reducing condensation at the cells considerably.

The position at which the gas is injected into the system is also important as the gas will be cold and dry.

When considering which parts of a rebreather will naturally condense and which will not, remember that breathing bags are a large surface area in contact with cold water. So a counter-lung is a very effective condensing device and has the effect of reducing the temperature of the loop gas. If your inhale counter-lung is in the direct gas path to the cells, you could consider insulating it. This is why neoprene counter-lungs can be so effective, but their construction (compressible at depth) makes them difficult to manufacture right.

Another school of thought may be to alter the temperature of the gas artificially. Heating the gas will have a similar effect to the scrubber itself, but will have the added advantage of increasing the efficiency of the scrubber, especially if it is a scrubber of average efficiency to begin with.

The oxygen injection point may have a reasonable effect of the sensor performance in relation to condensing water vapor  Purging instruments and systems to remove unwanted moisture is a standard practice. The method normally adopted is to allow dry gas to flow past the sensors for a defined period of time using a known dry gas. The dry gas in a rebreather can be either the oxygen or the diluent. This would constitute a similar advantage to single point purging. Although the injected gas would not completely purge the sensor as it will be mixed with loop gas, it may have a reasonable effect of lowering the total water vapor loading at the sensor face. In conclusion, there are potential advantages to the injection point being close to the sensors in opposition to other locations in respect to condensation.

Temperature Compensation

Temperature affects an oxygen cells output

We can get an extra burst of energy from a battery if we heat it up. This is because the heat causes the chemical reaction in the battery to increase and create more power. An oxygen cell reacts to temperature in a similar way.

Temperature compensation is VERY important

In an oxygen cell without temperature compensation, an increase of temperature from 20 degC to 40 degC can increase the current output from the cell by 50%. In terms of accuracy, the change in output from the cell from temperature swings has been stated to be as much as 2.5% per degC. So the effect of temperature is most significant.

Temperature compensate the reaction, not the gas

Oxygen cells are fitted with a compensation circuit to ensure this change does not affect its accuracy. What we should remember at this point is that we are compensating for the temperature of the reaction, not the temperature of the gas being sensed.
If you leave the oxygen sensor in the gas being sensed, eventually the sensor will reach equilibrium in temperature to the gas to which it is being subject to, but in the mean time, any compensation related to the gas temperature and not the reaction will create an error which may be significant.

Temperature changes in rebreathers are frequent and significant.

The temperature of a gas is directly affected by pressure, so we would expect that significant temperature swings will be expected by increasing pressure, like a decent or ascent on a dive. The frequent injection of oxygen and diluent and the continuous gas flow through the system negate this effect in practice. However, factors such as the large quantities of diluent injected during decent, calibrating with tank gas, the effect of the human breathing into the loop or the exothermic reaction of the scrubber will cause variance during operation. Temperature is important. In a rebreather, temperature swings are to be expected.

Validate the sensor by performing the right tests

If the tests for accuracy do not include temperature swings, and the system has been kept at a fairly constant temperature, then any issues with the compensation method will be unlikely to be found when sampling the gas against the control system.

Some oxygen cells do it differently

Some cells stated for diving use have specifications attached to them such as up to 15 mins to stabilize the signal output to the temperature compensation network.

One of the improvements of some cells is the temperature compensation specification. By referencing the compensation to the reaction rather than the gas being sensed, we can dramatically increase the accuracy of the cell.

At present all different manufacturers sensors will react differently to temperature. Other temperature compensation circuits will cause other cells to react differently and be voted out. Therefore it is our opinion that different manufacturer’s sensors MUST NOT be mixed.

Comments are closed.